STANFORD GRADUATE SCHOOL OF BUSINESS — California Governor Earl Warren, a relatively liberal Republican and future U.S. Supreme Court Chief Justice, was facing a difficult reelection challenge in 1950, the height of the McCarthy Era. With his conservative opponent making an issue of alleged Communist influence in California, Warren proposed the Levering Act, which required all state employees to take a loyalty oath and swear that they were not members of the Communist party.
Warren’s action appears surprising because only seven months earlier he had denounced a move to require the faculty of the University of California to sign loyalty oaths. What changed?
An election. With conservative Democrat James Roosevelt making an issue of Communist infiltration, Warren championed the act, even though his later writings show that he knew full well it was a violation of the spirit of the First Amendment and was likely to be ineffective as well, according to Kenneth Shotts, the David S. and Ann M. Barlow Professor of Political Economy at the Stanford Graduate School of Business. What’s more, opinion polls at the time showed that the public was frightened and in favor of anti-Communist legislation. And Warren must have known that.
Shotts has studied political decision making and leadership for over a decade. He wrote about Warren in a 2001 paper that identified and defined the contrast between leadership and pandering. “Elections sometimes give policymakers incentives to pander; i.e., to implement a policy that voters think is in their best interest, even though the policymaker knows that a different policy is actually better for the voters.” And that, of course, is just what Warren did.
In 2004, Shotts and Brandice Canes-Wrone, then an associate professor of political science at Northwestern University, studied presidential responsiveness over time. They examined 235 presidential budgetary proposals between 1972 and 1999. That research gave Shotts more insight into decision making, but it did not take into account the influence of the media.
Building on these two previous papers, Shotts and Scott Ashworth, of the University of Chicago, analyzed the common assumption that a healthy media would make office holders less likely to pander. To do this, they constructed a theoretical model using well-established principles of game theory and found some surprising results.
“By and large an effective media does make it less likely that politicians will pander. And it’s generally not a good thing to have the media play yes man,” says Shotts. “But there are some exceptions to this general rule,” he adds.
If the media always produced accurate commentary on policy choices, there would be less incentive for politicians to pander since voters would know what policies were in their interest.
However, in some situations an attentive media actually increases the likelihood of pandering. That’s because in both the real world and the theoretical model built by the researchers, the media is imperfect. Even when it is unbiased, it will make mistakes. In these cases, a politician who is considering an unpopular position that he or she believes to be good policy may fear that the media will mistakenly criticize that position and cause his defeat. Rather than face that possibility, the politician may opt for a position he knows to be against the interests of the voters.
Surprisingly, it also turns out that there are times when an office holder who knows that the media is almost always on his side (that is, acts as a yes man) believes he or she doesn’t have to worry that his policy choices will be criticized. At times, that freedom allows the politician to avoid pandering and take actions that are good for the voters without fear of being criticized by the media.
The paper, with these findings, “Does informative media commentary reduce politicians’ incentives to pander?” was written by Ashworth and Shotts, and published in the December 2010, issue of The Journal of Public Economics.
The model assumes that media commentators are unbiased and strive to present the news factually and that people act rationally in their best interests. Because these assumptions represent a best-case scenario for the media to reduce pandering, it is surprising that the media sometimes makes matters worse.
Even when well-meaning, the media sometimes acts as a yes man: “A yes-man media [is] one unwilling to criticize the incumbent unless it observes overwhelmingly clear information that the incumbent chose the wrong policy.” You’d think that was a bad thing, but again, the answer is not always. Because voters and the office holders are aware that the yes-man media will only criticize an office holder when he or she has very clearly made an error, the office holder is free to make what he thinks is the right policy choice without worrying about media criticism, a counter-intuitive finding that surprised the researchers.
Yes men aside, why is our intuitive belief that “good” media will lessen the incentive to pander sometimes mistaken? In the Ashworth and Shotts model, voters generally don’t judge events and policy choices symmetrically. “They reelect an incumbent who chooses a popular action unless he is proved wrong, but they only reelect an incumbent who chooses an unpopular action if he is proved right.”
If the media always informed the public accurately about policy choices, the asymmetry wouldn’t matter. But it doesn’t always get it right. “This asymmetric burden of proof creates the incentive for pandering — if the public is sufficiently unlikely to learn whether the incumbent’s policy choice was correct, then choosing the action with a lower burden of proof is optimal, even when that action is unlikely to be correct.” Or, to put it more simply: When in doubt, do the popular thing.
But don’t opponents in an election influence what an office holder might do? Does the fear of criticism from a challenger make for better — or worse — decisions by incumbents? Those are the questions that Shotts and Ashworth are exploring in their current research projects.
— Bill Snyder
Also on Stanford Knowledgebase: